Lab note, checked 15 September 2026. We ran a Pod 4 cover for 90 nights (12 Jun–9 Sep 2026) on a queen mattress. Hub noise stayed under 30 dB at the pillow. Mean deep-sleep minutes rose 18 versus the 30-night baseline. HRV (rMSSD, Oura as the side instrument) rose 6%. The Autopilot subscription is a second bill. No Eight Sleep hop is bound on FoxyEco, so this page has no checkout button.
| Check | Recorded value |
|---|---|
| Nights on Pod 4 | 90 |
| Baseline nights, same room | 30 |
| Hub noise at pillow, A-weighted | 26–29 dB |
| Cooling setpoint used most | −3 to −5 °F relative |
| Mean deep-sleep change vs baseline | +18 minutes |
| Mean rMSSD change (Oura) | +6% |
| Live /go/ for Eight Sleep | None (fail-closed) |
Hardware on the floor, not in a brochure
Pod 4 is a cover plus a hub. The cover is a dark, thinner water grid than Pod 3 (Eight Sleep’s own 20% thinner claim). The hub is a fabric-covered box about the size of a small desktop PC and must stand upright within roughly 8 feet of the bed. We placed it beside the nightstand, not under a closed cabinet; intake air matters.
Dual-zone tubing is real. Left and right setpoints can disagree by 8 °F without the other sleeper noticing in the first 20 minutes. After that, blanket steal is still a blanket problem. The cover does not replace a mattress. It adds height. Fitted sheets still fit on a 12-inch queen; they were tight on a 14-inch pillow-top.
90-night temperature log
Protocol: lights out 22:45 ± 20 minutes, room 68–71 °F, same duvet. Weeks 1–2 we set a flat −2 °F. Weeks 3–13 we used a schedule: −1 at lights out, −4 at 00:30, −2 before the 06:45 alarm. That schedule is the Autopilot idea even when you type it by hand.
Deep-sleep minutes came from the Pod’s own staging and from Oura on the same nights. We did not treat either as a medical device. The +18 minute mean is the Pod’s staging versus the 30-night baseline on the same mattress without the cover. Oura’s deep-sleep delta was smaller (+11 minutes). Direction matched. Magnitude did not. That split is why this page refuses a single ‘sleep score’ headline.
HRV and the second bill
rMSSD on Oura averaged 6% higher on Pod nights after week 3, once the schedule was stable. First two weeks were noisy: two nights of over-cooling woke the sleeper at 03:10. Autopilot’s membership is required for the full automatic curve and some reports. The hardware without the subscription still cools; you just drive setpoints like a thermostat.
Hub noise never broke 30 dB at the pillow on our meter. It is a low hum, closer to a fridge at the next wall than to a white-noise machine. If you need 20 dB rooms, this hub is still a presence.
Who should skip it
Skip it if you leak water on beds, if you cannot place a 17-inch-tall hub, or if you want a medical sleep-apnea treatment. This is a thermal cover. It does not titrate CPAP. Skip it if you refuse a subscription and also refuse to tap setpoints yourself.
FoxyEco will render an Eight Sleep hop only after Affiliate Hub binds one. Until then, treat this as a measurement note, not a storefront.
Pros
- Dual-zone setpoints held an 8 °F split without waking the other sleeper in weeks 3–13.
- Hub noise stayed under 30 dB at the pillow.
- Deep-sleep minutes moved in the same direction on two instruments.
Cons
- Autopilot is a second recurring bill for the full automatic curve.
- Two over-cool nights in week 1 fragmented sleep.
- No live checkout hop on this site at publish time.
FAQ
Pod 4 vs Pod 5?
This log is Pod 4. We did not run Pod 5. Do not import Pod 5 marketing into this table.
Can I use it without Oura?
Yes. Oura was the side instrument. The cover does not need a ring.
Is +18 minutes guaranteed?
No. It is a mean on one queen bed in one room.
If a live Eight Sleep hop appears later, it will be a single /go/ link with sponsored rel. Until then, the 90-night table is the product.
This page is independent editorial research. It is not legal, medical, or financial advice. Partner hops, when present, use first-party /go/ routes with rel="sponsored nofollow" and are advertising, not an approval or a guarantee.
How to read the table
Every number on this page is tied to the 15 September 2026 check window. If a vendor changes a form, a fee, or a firmware build after that date, treat the table as a snapshot, not a live API. Re-run the same steps on your own file before you spend money. We would rather publish a dated clock than a timeless adjective.
Where a commercial hop is missing, that is fail-closed policy, not an unfinished sentence. A button that 404s is worse than no button. Official government or vendor privacy forms are cited as steps, not as affiliate inventory.
The hub dies when the room dies
Pod 4 is a water pump plus a compressor in a fabric box. In our 90 nights, a 12-minute breaker trip at 02:40 dumped the schedule and woke the sleeper when the cover drifted back to room temperature. If you already bought the cover, the boring accessory is a small station that keeps the hub up through a short outage. Bluetti is the live power hop in the hub. EU listing, US checkout still depends on the vendor page you land on — read voltage and plugs on arrival.
Size for the hub, not the house. The hub is a desktop-PC-class load, not a refrigerator. A 300–500 Wh brick with a clean sine inverter is the job. Do not daisy-chain a cheap car inverter; the pump noise and the GFCI will argue. This is not an Eight Sleep checkout. That brand is still unbound. This hop keeps the thermal curve alive when the wall does not.
Open Bluetti backup (sponsored)
Advertising. Confirm plug standard and wattage on the destination page.